Sage Journals: Discover world-class research

Abstract

The authors investigated public support for government repression of protests (police repression, legal repression, and punishment of protesters) following incidents of violence and harm. Using two factorial vignette experiments embedded in a national Qualtrics survey (n = 1,229), the authors examined whether partisan bias (i.e., polarized responses to actions by ideological opponents or allies) characterized public preferences for repressive government responses to intentional violence (i.e., rock throwing) or incidental harm (i.e., coronavirus disease 2019 transmission) occurring at protests. The authors also examined whether violence or harm severity, or violence against or harm to police, influenced the degree of partisan bias in public responses. The results indicated partisan bias in support of police repression and punishment preferences and, to a lesser extent, legal repression. Members of the public preferred more repressive responses to political opponents and less repressive responses to political allies. Partisan bias in preferences for punishment was also heightened when a police officer was the target of intentional violence.

Keywords

harm to police political ideology protest violence punitiveness repression

Protest movements have been at the forefront of scholarly and public discourse in recent years (Cobbina 2019; Metcalfe and Pickett 2022; Olzak 2021). In the summer of 2020, millions of people marched in support of the Black Lives Matter (BLM) movement, protesting systemic state violence against Black communities. Despite the overwhelmingly nonviolent nature of the movement, violence and property destruction occurred in some locations (Chenoweth and Pressman 2020; Cho et al. 2022). The same year also saw collective action from the political right, including “make America great again” (MAGA) protests aimed at securing a second presidential term for Donald Trump. This movement culminated in the January 6, 2021, insurrection during which five people died, including a police officer (Kishi, Stall, and Jones 2020; Slobin and Hart 2021). Although BLM and MAGA are not equivalent as social movements, nor was the violence associated with each movement the same (Brantley-Jones 2021), public and media discourse about both illustrated the salience of violence and harm in discussions about collective action. Whereas the vast majority of both left- and right-leaning protests in the United States were nonviolent throughout 2020 and 2021 (Kishi and Jones 2020; Kishi et al. 2021), violence and harm by protesters consistently made headlines (Cho et al. 2022; Drakulich and Denver 2022; Feinberg, Willer, and Kovacheff 2020; Gutting 2020), a common theme in media coverage of protest movements (e.g., Edwards and Arnon 2021).

Public perceptions of violence by protesters, and beliefs about appropriate responses to protest violence, are also important. Broadly, public perceptions of social movements may shape political and social structures that facilitate or inhibit social change (e.g., Bugden 2020; Heaney and Rojas 2011; Mische 2009). People tend to support protest movements they perceive as nonviolent (Bugden 2020; Edwards and Arnon 2021; Gutting 2020; Stephan and Chenoweth 2008; Thomas and Louis 2014). Public support for repression of protests may likewise signal to policy makers what responses to protesters are politically viable (Baranauskas 2022; Goff, Silver, and Iceland 2022; Hsiao and Radnitz 2021; Metcalfe and Pickett 2022).

The present study adds to this literature by exploring public support for repressive government responses to violence and harm occurring at protests. Specifically, we explore how responses to violence and harm at protests are conditioned by the extent to which observers agree with the protest cause (Edwards and Arnon 2021; Feinberg et al. 2020; Hsiao and Radnitz 2021). Research suggests that people perceive their ideological opponents as more violent and deserving of repression than their ideological allies, even when those groups have engaged in identical protest actions (Barker, Nalder, and Newham 2021; Edwards and Arnon 2021; Feinberg et al. 2020; Gutting 2020; Hsiao and Radnitz 2021). However, little is known about how such partisan bias characterizes support for different forms of repression, or how partisan bias varies across contexts (Hsiao and Radnitz 2021). In the present study we consider partisan bias in public support for police and legal repression of protest movements and punishment of individual protesters, as well as how situational characteristics (e.g., the severity of the violence or harm and whether police officers or civilians were affected) may shape partisan bias.

As noted previously, it is crucial for advocates to understand what shapes public responses to protests, including instances of violence or harm that are overrepresented in media coverage, because public support for causes and resistance toward government repression are needed for protest movements to thrive (Baranauskas 2022; Edwards and Arnon 2021; Feinberg et al. 2020; Heaney and Rojas 2011; Hsiao and Radnitz 2021). Understanding support for punishing individual protesters is also important because fear of legal repercussions suppresses participation in protests (Earl 2011) and characterizations of protesters as “criminal” detract from the legitimacy of protest movements (Baggetta and Meyers 2022). More broadly, the study sheds light on the complex ways in which partisanship may condition preferences for the social control of ideological allies and opponents (e.g., Mische 2009), which is increasingly important given heightened political “tribalism” in recent years (e.g., Iyengar and Westwood 2015).

We examine partisan bias in support of the repression of protests using factorial vignette experiments embedded in a Qualtrics Panel (QP) survey (n = 1,229). The vignettes described violence (rock throwing) and harm (coronavirus disease 2019 [COVID-19] transmission) occurring during protests aligned with the political left (BLM) or political right (MAGA); manipulations included violence and harm severity and whether a police officer or civilian was a victim. Partisanship was measured using self-identification on a conservative-liberal continuum.

Public Responses to Protest Violence and Harm

Numerous studies suggest that violence by protesters reduces the perceived legitimacy of, and identification with, protest movements (e.g., Chenoweth, Stephan, and Stephan 2011; Edwards and Arnon 2021; Feinberg et al. 2020; Huff and Kruszewska 2016; Simpson, Willer, and Feinberg 2018; Thomas and Louis 2014; but see Shuman et al. 2022). Violence may also play a role in shaping support for government repression of protest movements (Edwards and Arnon 2021; Goff et al. 2022; Metcalfe and Pickett 2022). Defined by social movement scholars as “any action by another group which raises the contender’s cost of collective action” (Tilly 1978:100), repression may include “law-and-order” policing of protesters (e.g., Goff et al. 2022), legal bans on protest activities (e.g., Davenport 2007), or punishing individuals who engage in protests (e.g., Barkan 2006), among other responses. Broadly, the public appears more willing to accept the government repression of violent protests compared with nonviolent protests (Edwards and Arnon 2021; Goff et al. 2022; Hsiao and Radnitz 2021; Metcalfe and Pickett 2022).

Partisanship also matters. Although social movements are often characterized as working “outside the system,” the boundaries between social movements and political parties are often porous (Heaney and Rojas 2011). Broadly, alignment with political parties brings benefits (e.g., motivation of support and access to resources) and drawbacks (e.g., infighting and divisiveness) to social movements (Mische 2009). At an individual level, partisan identity may shape receptiveness to peaceful protest messaging (Bugden 2020).

Less is known, however, about the role of individual differences, including partisan identity, in shaping public responses to instances of violence or harm occurring at protests (Feinberg et al. 2020). This is an important oversight because ideological identification informs interpretations of facts (Kahan et al. 2017) and attributions of blame (Bisgaard 2015). People with opposing ideological alignments may therefore perceive the same incident differently (Hsiao and Radnitz 2021; Feinberg et al. 2020; Iyengar and Westwood 2015), “judging their opponents harshly while giving seemingly allied groups the benefit of the doubt” (Hsiao and Radnitz 2021:481). In turn, they may differ in their support for repressive responses to protest violence by ideological allies and opponents. Such partisan bias may be especially pronounced in responses to protests, which are inherently conflictual and polarizing across ideological lines (Feinberg et al. 2020; Hsiao and Radnitz 2021).

A few studies suggest that this is the case.¹ Hsiao and Radnitz (2021) found that Republicans viewed left-leaning protests as more violent than did Democrats, while members of both parties viewed right-leaning protests similarly, with downstream effects on support for arresting organizers and enacting punitive antiprotest laws. Edwards and Arnon (2021) found that while people did not perceive ideologically opposed protesters as more violent, they nevertheless supported more repressive measures against them. Giersch (2019) found that college students supported more criminal punishment for ideologically opposed campus protesters.

However, gaps in the literature remain. Only a few studies have examined partisan bias in support of repressive measures and largely have not compared support for different repressive measures. Research has also focused exclusively on intentional violence or disruption but has neglected other forms of harm associated with protests; we consider both intentional violence (rock throwing) and incidental harm (COVID-19 transmission). Additionally, little work has explored situational characteristics that may affect whether people evaluate protest violence or harm through a partisan lens. To address this latter gap in the research, we consider two potential sources of partisan bias in response to protest violence: the severity of the violence or harm, and whether police officers or civilians suffered violence or harm.

Severity of Violence or Harm

A situational characteristic that may shape the degree of partisan bias in support for repression of protests is the severity of the violence or harm. More severe infractions (e.g., those that are more harmful or involve intentional disregard for others’ wellbeing) may be difficult to rationalize, even by ideological allies. Less severe forms of violence or harm may be open to partisan interpretation. For example, Hsiao and Radnitz (2021) found stronger partisan biases in perceptions of the harmfulness of nonviolent disruption (e.g., chanting with placards or blocking a highway) than of throwing rocks (see also Simpson et al. 2018).

However, research has largely focused on the use of violent vs. nonviolent tactics, rather than gradations in the severity of violence or harm, a potentially important distinction given the salience of violence and harm in the public discourse. In the present study, we consider both harm and intent as sources of perceived severity (e.g., Darley 2009). We predict that partisan bias will be reduced for more severe violence or harm (i.e., violence resulting in injury or harm caused through intentional disregard for others) but increased for less severe violence or harm (i.e., violence with no injury or unanticipated harm).

Violence against (or Harm) to Police

Another situational characteristic that may condition partisan bias in support for repression is whether police officers or civilians² suffer violence or harm. Public views on violence against the police can be variable (Maguire et al. 2020; Tyler et al. 2018). Police are omnipresent at protests, and antipolice violence may be touted by ideological opponents as evidence that a protest movement is out of control (e.g., Irvine 2020). However, because police have long been used to quell social unrest (e.g., Drakulich and Denver 2022), antipolice violence may also be characterized as justifiable or even desirable by ideological allies (Maguire et al. 2020). Police are also the most visible representatives of the state (e.g., Bradford, Murphy, and Jackson 2014), which protests typically challenge. Violence toward police may therefore be perceived by allies as a legitimate expression of grievance against the state or, by opponents, as an unacceptable affront to the state’s authority (Gutting 2020). In contrast, violence against or harm to civilians may have little symbolic value and result in condemnation from both sides. We predict that partisan bias in support for repression will be stronger when a police officer suffers violence or harm, and weaker when a civilian suffers violence or harm.

Asymmetry in Partisan Bias

Are conservatives or liberals more biased in their evaluation of protests? This question does not yet have a clear answer, both because few studies have considered asymmetry in partisan responses to protests (Barker et al. 2021) and because extant findings are mixed. Hsiao and Radnitz’s (2021) findings suggest that Democrats’ ratings of the harmfulness of protest violence varied more by protest cause than did Republicans’ ratings. Barker et al. (2021) similarly found that liberals viewed left-leaning protests much more favorably than right-leaning protests, whereas conservatives viewed right-leaning protests only somewhat more favorably than left-leaning protests. In contrast, Feinberg et al. (2020) found little evidence of asymmetry in support for protest movements, consistent with research suggesting that partisan biases tend to be largely symmetrical outside of the protest context (e.g., Ditto et al. 2019). In the present study, we therefore consider, but do not test specific hypotheses regarding, asymmetry in partisan biases among liberals and conservatives.

The Present Study

The present study aims to contribute to the literature in a few ways. First, we expand on earlier research by examining support for differing government responses to protest violence or harm. Second, we examine the role of situational characteristics, including the severity of violence and harm and whether police are recipients of violence and harm, in shaping those responses. Third, we explore these relationships in scenarios involving intentional violence (i.e., rock throwing) and incidental harm (i.e., COVID-19 transmission). Broadly, we aim to shed light on how people respond to perceived infractions committed by members of ideologically aligned or opposing political “tribes.” To this end, we test the following three hypotheses:

Hypothesis 1: Political identification will interact with protest cause so that individuals more strongly support repressive responses to ideologically opposed protesters and more weakly support repressive responses to ideologically aligned protesters.

Hypothesis 2: Partisan bias will be stronger when a violent or harmful action is less severe (i.e., less harmful or intentional), and weaker when a violent or harmful action is more severe.

Hypothesis 3: Partisan bias will be stronger when a police officer experiences violence or harm, and weaker when a civilian experiences violence or harm.

Data

Data were drawn from a national survey conducted by QP in 2021. QP sends out online survey invitations to opt-in panelists, who receive small payments (about $2), to achieve quota sampling. This method produces demographic and politically representative samples (Boas, Christenson, and Glick 2020) and typically provides higher quality data than online crowdsourcing platforms such as Amazon Mechanical Turk (Chandler et al. 2019; Zack, Kennedy, and Long 2019). Our sample was populated to be representative of the United States on gender, race, age, and region. Institutional review board approval was obtained prior to conducting the study and all participants indicated their willingness to participate in the study via an informed consent document.

In all, 1,260 respondents received compensation for completing the survey. The analytic sample included respondents who completed all relevant items (n = 1,229). Of these, 49 percent of respondents were male (compared with 50 percent of the U.S. population; see U.S. Census Bureau 2022), 69 percent were white (compared with 76 percent of the U.S. population); 17 percent were Hispanic/Latino (compared with 19 percent of the U.S. population), and 20 percent of the sample was older than 65 years (compared with 22 percent of adults in the U.S. population). The median household income was between $40,000 and $49,999 (compared with $64,994 in the U.S. population) and 37 percent of individuals aged 25 and older held a bachelor’s degree (compared with 33 percent in the U.S. population). There was also variation in ideological views: 28 percent of respondents were liberal, 42 percent were moderate, and 30 percent were conservative; this distribution is similar to 2021 General Social Survey (https://gss.norc.org) estimates of political identification in the United States (33 percent liberal, 35 percent moderate, 32 percent conservative).

The survey included an attention check, which 896 respondents (71.11 percent) passed (of these, 887 answered all relevant questions). The attention check was adapted from Berinsky, Margolis, and Sances (2014) and included the following text:

In surveys like ours, some participants do not read all the questions carefully. To show that you are reading the questions carefully, please select Fox News and MSNBC from the list below. What sources do you use to learn about crime in the news?

Respondents were then presented with a list of news sources (CNN, Fox News, MSNBC, CBS, and NBC News). Although we retain the full sample in the main analysis to preserve the demographic diversity of the sample (see Berinsky et al. 2014), we also conducted sensitivity analyses among only the subset of respondents who passed the attention check (see Appendix A, Table A1). All results were substantively similar. However, the political identity × cause interaction effects were larger among attentive respondents than in the full sample, suggesting that the main results may somewhat understate the degree of partisan bias.

Materials and Measures

We use data from two experimental vignettes in which protesters engaged in violent or harmful actions at a left-leaning (i.e., BLM) or right-leaning (i.e., MAGA) protest event. Vignette 1 described a protester throwing a rock at an observer, while vignette 2 described a protester transmitting a fatal case of COVID-19. Because the scenarios differ in various ways (e.g., the type of action, whether the harmful action is a criminal offense, and the ultimate outcome), similar findings may provide confidence that the results are not idiosyncratic to a particular scenario. Both vignettes also involve forms of violence and harm that have drawn public attention in recent years: rock throwing (particularly at police officers) has been widely discussed (Cho et al. 2022), and COVID-19 spread constituted a concern during pandemic-era protesting (although actual rates of transmission were low; Cobbina et al. 2021). Both vignettes also included experimental manipulations for (1) the severity of the violent or harmful action and (2) whether a police officer or a civilian suffered violence or harm. Table 1 provides an overview of the experimental conditions that were presented to respondents.

Table 1.

Experimental Design.

Manipulation(s)	Experimental Conditions
Protest cause (2 conditions, randomized)	BLM protest		MAGA protest
Vignette 1 (4 conditions, randomized)	Injury/police victim	Injury/civilian victim	No injury/police victim	No injury/civilian victim
Vignette 2 (4 conditions, randomized)	Positive test/police victim	Positive test/civilian victim	False-negative test/police victim	False-negative test/civilian victim

Note: Text containing each manipulation or set of manipulations appeared on a separate page of the survey. BLM = Black Lives Matter; MAGA = make America great again.

Prior to reading the vignettes, respondents were randomly assigned to read a statement varying the protest cause: half were told that the “protesters’ main goal was to address systemic racism in the United States” (reflecting a common theme in BLM protest events), while the other half were told that the “protesters’ main goal was to return the former President Trump to office” (reflecting a common theme in MAGA protest events). This information was randomized prior to the presentation of both vignettes so that all respondents read two vignettes about the same protest event. Given that media coverage of protest violence usually focuses on one protest event at a time, we sought to avoid having respondents make direct comparisons between the two protest causes. This is important because survey respondents typically attempt to provide consistent answers to questions, including evaluating “both sides” more similarly if they are presented together (Bradburn, Sudman, and Wansink 2004).

The protest causes (i.e., BLM and MAGA), which were intended to elicit perceptions of the protesters as ideological allies or opponents, were chosen for their salience at the time of data collection (i.e., summer of 2021) and the extent to which support for each movement was politically polarized. In 2021, 85 percent of Democrats supported the BLM movement, whereas only 19 percent of Republicans did (and only 4 percent indicated strong support; Horowitz 2021). Republicans in 2021 were more likely to believe that the 2020 election was “stolen” from Trump (81 percent of Republicans vs. 3 percent of Democrats) and that the storming of the U.S. Capitol on January 6, 2021, which arose from a MAGA protest, was a legitimate protest (47 percent of Republicans vs. 14 percent of Democrats) (Blake 2022).³ We anticipated that liberal respondents would perceive BLM protesters as ideological allies, whereas conservative respondents would perceive MAGA protesters as ideological allies, and that each would view the other as opponents; a similar approach has been used in prior work (e.g., Feinberg et al. 2020; Hsiao and Radnitz 2021). In subsequent analyses, protest cause is coded so that 1 = BLM protest and 0 = MAGA protest.

Vignette 1

As noted above, vignette 1 described a scenario in which a protester threw a rock at a person. Manipulations were presented using a 2 × 2 factorial design following the initial assignment to protest cause. The recipient of the violence was either a police officer or a bystander (the police victim manipulation, coded 1 = police officer and 0 = bystander). The rock either injured or missed the individual (the injury manipulation, corresponding to greater infraction severity; this is coded 1 = injury and 0 = no injury). The full text⁴ is as follows:

Jerry is a 40-year-old man who attended the protest. During the protest, Jerry threw a rock at a 32-year-old [police officer/ bystander] named Aaron who was observing the protest. The rock [hit Aaron in the head, giving him a bad concussion/missed Aaron and didn’t hurt him at all].

Vignette 2

Vignette 2 described a scenario in which a protester transmitted COVID-19 to another individual, who ultimately died. The manipulations were also presented using a 2 × 2 factorial design. The individual who died was either a police officer or an emergency medical technician (EMT) (the police victim manipulation, coded 1 = police officer and 0 = EMT). The protester had either received a positive test result for COVID-19 or a false negative test result for COVID-19 (the positive test manipulation, in which more intentional disregard for others corresponds to greater infraction severity; this is coded 1 = positive test and 0 = false-negative test). The full text is as follows:

Mike is a 29-year-old man who attended the protest. At the time of the protest, Mike was sick with COVID-19 and had cold-like symptoms. He had received a [positive test result for COVID-19 the day before/false negative test result for COVID-19 the day before, meaning he was not aware that he had COVID-19]. During the protest, Mike infected a 53-year-old [police officer/emergency medical technician (EMT)] named William who was stationed at the protest. William later died from COVID-19 complications.

Dependent Variables

Following each vignette, respondents were asked to imagine they “heard about what Jerry (Mike) did” and indicate their support for government actions regarding “protests with similar goals to the protest Jerry (Mike) was involved in.” Items measuring support for police repression included “using police with riot gear to control future protests” and “using the military to control future protests” (1 = “strongly oppose” to 5 = “strongly support”); responses were averaged to create scales (α = .817 for vignette 1, α = .858 for vignette 2). Items measuring support for legal repression included “banning future protests altogether” and “banning any kind of political action by groups with this goal” (1 = “strongly oppose” to 5 = “strongly support”); responses were averaged to create scales (α = .837 for vignette 1, α = .891 for vignette 2).⁵ Measuring punishment preferences, respondents indicated what punishment, if any, they thought appropriate for Mike or Jerry (1 = “no punishment,” 2 = “a fine,” 3 = “probation,” 4 = “prison—less than 5 years,” and 5 = “prison—5 years or more”).

Independent Variables

Political identity was measured on a liberal-conservative continuum (1 = “very liberal,” 2 = “somewhat liberal,” 3 = “moderate” (middle of the road), 4 = “somewhat conservative,” and 5 = “very conservative”). In subsequent analyses, we examine whether variation on the scale (i.e., lower scores indicating greater liberalism or higher scores indicating greater conservatism) is associated with support for repressive responses to BLM and MAGA protests. However, given the limitations of a continuum-based approach for understanding political “tribalism” (e.g., the treatment of moderates as a neutral midpoint; see Fowler et al. 2023), we also compare views held by respondents at each end of the scale (i.e., strong liberals and strong conservatives). Note that supplemental analyses using an alternate coding scheme for the political identity variable (i.e., recoding the continuum as a nominal-level measure of liberal, conservative, or moderate identity) also returned similar results.

Control variables relevant to perceptions of BLM and MAGA protests included racial resentment and having voted for President Trump. Support for Trump was measured using an item asking what candidate (if any) respondents had voted for in the 2020 Presidential election; responses were coded so that 1 = voted for Trump and 0 = did not vote for Trump. To measure racial resentment, respondents were asked to indicate their agreement with the following items (see Enns and Ramirez 2018): “Irish, Italians, Jewish, and many other minorities overcame prejudice and worked their way up. Blacks should do the same without any special favors”; “Generations of slavery and discrimination have created conditions that make it difficult for Blacks to work their way out of the lower class” (reversed); “The Irish, Italians, Jews and many other ethnic groups immigrated to the United States legally. Latinos and Hispanics should do the same without any special favors”; and “Latinos and Hispanics would be more welcome in the United States if they would try harder to learn English and adopt U.S. customs like past immigrant groups have done” (1 = “strongly disagree” to 5 = “strongly agree”); responses were averaged to form a scale (α = .760).

Additional control variables included participation in protests, experiences with police, and demographics. Prior protest participation was measured as the number of protests respondents attended in the past five years (1 = “none,” 2 = “1 or 2,” 3 = “3 or 4,” and 4 = “5 or more”). Respondents also characterized their experiences with police in the past five years as “mostly good,” “mixed—some were good, and some were bad,” or “mostly bad”; dummy variables were created for positive police experience, mixed police experience, and negative police experience (the reference category was no experiences with police).

Gender was coded so that 1 = male and 0 = female or nonbinary.⁶ Race was coded using dummy variables for Black and other race (the reference category was white). Ethnicity was coded so that 1 = Hispanic/Latino and 0 = not Hispanic/Latino. Age was coded in years from 18 to 95. Education was coded from 1 = less than high school to 7 = doctorate. Income was coded from 1 = less than $10,000 to 12 = more than $150,000. Region was coded using dummy variables for the South, Midwest, and West (the reference category was Northeast). Multicollinearity diagnostics indicated that variance inflation factors for all independent variables were ≤2.0. Descriptive statistics appear in Table 2.

Table 2.

Descriptive Statistics (n = 1,229).

Variables	Mean	SD	Range
Vignette 1, police repression	3.42	1.21	1–5
Vignette 1, legal repression	2.75	1.31	1–5
Vignette 1, punishment	2.84	1.10	1–5
Vignette 2, police repression	3.10	1.35	1–5
Vignette 2, legal suppression	2.64	1.39	1–5
Vignette 2, punishment	2.38	1.31	1–5
Political identity	3.04	1.16	1–5
Racial resentment	3.28	1.02	1–5
Voted for Trump	.31	—	0–1
Protest experience	1.34	.68	1–4
Positive police experience	.40	—	0–1
Mixed police experience	.24	—	0–1
Negative police experience	.08	—	0–1
Education	3.70	1.51	1–7
Income	5.48	3.31	1–12
Male	.48	—	0–1
Hispanic/Latino	.17	—	0–1
Black	.15	—	0–1
Other race	.16	—	0–1
Age	46.66	17.80	18–95
Midwest	.21	—	0–1
South	.37	—	0–1
West	.23	—	0–1

Note: Standard deviations are not reported for dummy variables.

Analysis Plan

We use ordinary least squares regression to predict each outcome (i.e., support for police repression, legal repression, and punishment) for each vignette. We present main effects, then interact political identity and protest cause (i.e., political identity × protest cause) to assess the degree of partisan bias along a liberal-conservative continuum.⁷ For each model including a two-way interaction term, we also compare the predicted mean levels of support for each form of repression among strong liberals and strong conservatives, who may better represent opposing political “tribes”; these comparisons also allow us to assess the degree of asymmetry in partisan bias among strong liberals and strong conservatives. Models including three-way interactions for political identity × protest cause × severity and political identity × protest cause × police victim are used to assess whether partisan bias differs across the severity and police victim conditions. Where relevant, adjusted predictions (with covariates set at their means) are presented to illustrate two- and three-way interactions.

Our analysis uses 18 regression models including two- or three-way interaction effects to test hypotheses 1 to 3 (i.e., 6 models examining two-way interactions and 12 models examining three-way interactions, not counting “baseline” models with no interactions or supplemental analyses). As such, we use a Bonferroni-adjusted significance level of .05/18 = .003 to evaluate our hypotheses. Coefficients with p values from .003 < p < .05 are interpreted as “suggestive” of an effect (see, e.g., Benjamin et al. 2018) but nonsignificant. For convenience, we note the adjusted significance level of p < .003 in tables presenting regression results.

Finally, to test the robustness of our results to alternate measures of partisanship, we also conducted supplemental analyses assessing partisan bias arising from support for President Trump (which may drive identification with or opposition to both BLM and MAGA protests; see Blake 2022; Drakulich et al. 2020). Finding partisan bias based on alignment with Trump would support the contention that bias reflects perceptions of protesters as ideological allies or opponents (rather than reflecting identification on a liberal-conservative continuum only).⁸

Results

Main Effects and Political Identity × Protest Cause Interactions (Vignette 1)

Table 3 shows regression models predicting support for police repression (models 1 and 2), legal repression (models 2 and 3), and punishment (models 4 and 5) in vignette 1 (rock throwing). The odd-numbered models show main effects for the experimental manipulations, political identity, and controls, and provide a baseline for interpreting subsequent results. The even-numbered models introduce the political identity × protest cause interactions.

Table 3.

Regression Models Predicting Support for Repressive Responses to Rock Throwing at Protest (Vignette 1; n = 1,229).

	DV: Police Repression				DV: Legal Repression				DV: Punish Protester
	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
Variable	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE
Political identity × Cause	—	—	.186**^†	.054	—	—	.179**	.062	—	—	.232***^†	.051
Manipulations
Cause (BLM vs. MAGA)	−.134*	.062	−.701***^†	.174	−.164*	.071	.049	.071	−.056	.059	−.763***^†	.166
Police victim	.201**^†	.062	.180**	.062	.069	.071	−.036	.071	.140	.059	.114	.059
Injury	.031	.062	.026	.062	−.031	.071	−.709***^†	.201	.698***^†	.059	.692***^†	.059
Political identity	.029	.031	−.065	.041	−.061	.036	−.151**^†	.048	−.021	.030	−.138***^†	.039
Racial resentment	.434***^†	.035	.429***^†	.035	.389***^†	.041	.384***^†	.041	.083*	.034	.076*	.034
Voted for Trump	.076	.080	.081	.080	.046	.093	.050	.092	−.071	.077	−.065	.076
Protest participation	.059	.050	.061	.050	.171**	.058	.173**	.057	.023	.048	.025	.048
Positive police experience	.212**	.078	.211**	.078	−.017	.090	−.017	.090	.053	.075	.053	.074
Mixed police experience	.073	.093	.070	.093	−.056	.107	−.058	.107	.015	.089	.012	.088
Negative police experience	−.348**	.132	−.343**	.132	−.382*	.152	−.377*	.152	−.062	.127	−.055	.126
Education	−.001	.024	−.001	.024	−.069*	.027	−.070*	.027	.008	.023	.007	.023
Income	.010	.011	.011	.011	−.018	.012	−.018	.012	.015	.010	.016	.010
Male	−.086	.066	−.083	.065	−.166*	.076	−.163*	.076	.129*	.063	.133*	.062
Hispanic/Latino	−.195*	.098	−.214*	.098	−.203*	.113	−.221	.113	−.043	.094	−.068	.094
Black	−.057	.099	−.058	.098	.030	.114	.029	.113	.062	.095	.061	.094
Other race	.058	.098	.079	.098	.095	.113	.115	.113	−.094	.094	−.068	.094
Age	.002	.002	.002	.002	−.006*	.003	−.006*	.003	.000	.002	.000	.002
South	.088	.099	.084	.099	−.013	.114	−.016	.114	−.006	.095	−.010	.094
Midwest	.114	.090	.108	.090	.095	.104	.090	.103	−.048	.086	−.055	.085
West	−.093	.098	−.078	.098	−.136	.113	−.122	.113	.036	.094	.054	.094
R ²	.215		.223		.119		.125		.127		.141

Note: b = unstandardized regression coefficient; BLM = Black Lives Matter; DV = dependent variable; MAGA = make America great again.

p < .05. **p < .01. ***p <. 001. ^†p < .003.

In the odd-numbered baseline models, respondents who read about a BLM protest (vs. a MAGA protest) were less supportive of police repression (model 1; b = −.134, p = .030) and legal repression (model 3; b = −.164, p = .022) but were not significantly less punitive toward the protester. The police victim condition predicted greater support for police repression (model 1; b = .201, p = .001), whereas the injury condition was associated with greater support for punishment (model 5; b = .698, p < .001). Political identity was not significantly associated with any outcome. Among the controls, racial resentment predicted greater support for police repression, legal repression, and punishment; protest participation predicted greater support for legal repression; negative police experiences predicted reduced support for police and legal repression; and male and older respondents were less supportive of legal repression, whereas Hispanic/Latino respondents were less supportive of police repression.

In the even-numbered models including interactions, political identity interacted with protest cause in the expected direction in predicting support for police repression (model 2; b = .186, p = .001), legal repression (model 4; b = .179, p = .004), and punishment (model 6; b = .232, p < .001); the effects for police repression and punishment were significant at the Bonferroni-adjusted significance level (p < .003), while the effect for legal repression was “suggestive” of a relationship only.

Figure 1 shows the predicted values of each outcome for the BLM and MAGA conditions. In the BLM condition, greater conservatism (reduced liberalism) was associated with greater support for police and legal repression and greater punitiveness. In the MAGA condition, greater conservatism (reduced liberalism) was associated with reduced support for police and legal repression and reduced punitiveness. At the end points of the continuum, for all outcomes, strong liberals and strong conservatives preferred more repressive responses to violence by ideological opponents than by ideological allies. The gap in these preferences was wider and more distinct for strong liberals than for strong conservatives in predicting support for police and legal repression, while the effects were more symmetrical in predicting punishment preferences.

Figure 1.

Support for repressive responses to protest violence by protest cause (vignette 1 [V1]): (a)support for police repression (V1), (b)support for legal repression (V1), and (c)punishment preferences (V1).

Comparing⁹ adjusted predictions at each scale end point supports this interpretation. Strong liberals supported significantly less police repression (difference = −.514, p < .001) and legal repression (differ-ence = −.529, p < .001) of BLM protesters, compared with MAGA protesters. In contrast, strong conservatives did not support significantly more police repression (difference = .231, p = .058) or legal repression (difference = .188, p = .181) of BLM protesters, compared with MAGA protesters. However, both strong liberals (difference = −.530, p < .001) and strong conservatives (difference = .400, p < .001) supported significantly harsher criminal punishment of the other side’s protesters.

Main Effects and Political Identity × Protest Cause Interactions (Vignette 2)

Table 4 shows regression models predicting support for police repression (models 1 and 2), legal repression (models 2 and 3), and punishment (models 4 and 5) in vignette 2 (COVID-19 transmission). The odd-numbered models again show main effects while the even-numbered models introduce the political identity × protest cause interactions.

Table 4.

Regression Models Predicting Support for Repressive Responses to Coronavirus Disease 2019 Transmission at Protest (Vignette 2; n = 1,229).

	DV: Police Repression				DV: Legal Repression				DV: Punish Protester
	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
Variable	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE
Political identity × cause	—	—	.251***^†	.062	—	—	.151*	.065	—	—	.220***^†	.060
Manipulations
Cause (BLM vs. MAGA)	−.234**^†	.072	−.999***^†	.201	−.191*	.076	−.650**^†	.212	−.002	.069	−.672**^†	.195
Police victim	.093	.072	.099	.071	.103	.075	.106	.075	.085	.069	.090	.069
Positive test	.228**^†	.072	.222**^†	.071	.154*	.075	.151*	.075	.970***^†	.069	.965***^†	.069
Political identity	−.033	.036	−.159**^†	.048	−.070	.038	−.146**	.050	−.035	.035	−.146***^†	.046
Racial resentment	.437***^†	.041	.430***^†	.041	.388***^†	.043	.384***^†	.043	.092*	.040	.085*	.039
Voted for Trump	.040	.093	.048	.093	.011	.098	.015	.098	−.073	.090	−.066	.090
Protest participation	.189**^†	.058	.192**^†	.058	.262***^†	.061	.264***^†	.061	.187**^†	.056	.190**^†	.056
Positive police experience	.147	.090	.146	.090	.051	.095	.051	.095	−.136	.087	−.137	.087
Mixed police experience	−.036	.108	−.039	.107	−.080	.113	−.081	.113	−.136	.104	−.138	.104
Negative police experience	−.253	.153	−.245	.152	−.300	.161	−.295	.161	−.058	.148	−.050	.147
Education	−.022	.028	−.023	.027	−.082**	.029	−.083**	.029	.016	.027	.014	.026
Income	.005	.012	.006	.012	−.031*	.013	−.030*	.013	−.011	.012	−.010	.012
Male	−.148	.076	−.144	.076	−.189*	.080	−.186*	.080	.078	.074	.081	.073
Hispanic/Latino	−.123	.114	−.148	.114	−.044	.120	−.059	.120	.218*	.110	.196	.110
Black	−.032	.115	−.034	.114	−.067	.120	−.069	.120	.146	.111	.145	.110
Other race	.155	.114	.183	.114	.087	.120	.104	.120	.279*	.110	.303**	.110
Age	.003	.003	.003	.003	−.007*	.003	−.007*	.003	.003	.003	.003	.003
South	−.063	.115	−.067	.114	−.089	.121	−.092	.121	−.059	.111	−.063	.110
Midwest	.068	.104	.061	.104	.070	.110	.066	.109	−.144	.101	−.150	.100
West	−.266*	.114	−.247*	.113	−.168	.120	−.156	.120	−.082	.110	−.065	.110
R ²	.156		.167		.126		.130		.168		.177

Note: b = unstandardized regression coefficient; BLM = Black Lives Matter; DV = dependent variable; MAGA = make America great again.

p < .05. **p < .01. ***p < .001. ^†p < .003.

In the odd-numbered baseline models, respondents again had less support for police repression (model 1; b = −.234, p = .001) and legal repression (model 3; b = −.191, p = .012) of BLM protests than MAGA protests, whereas punitiveness did not vary by protest cause. Respondents were more likely to support police repression (model 1; b = .228, p = .002), legal repression (model 3; b = .154, p = .041), and punishment (model 5; b = .970, p < .001) when the protester had tested positive for COVID-19. Neither the police manipulation nor political identity was significantly associated with any outcome. Among the control variables, racial resentment predicted greater support for all outcomes; protest participation predicted increased support for all outcomes; education and income predicted reduced support for legal repression; male and older respondents were less supportive of legal repression; “other” race respondents were more punitive (compared with white respondents); and residence in the West predicted less support for police repression (compared with residence in the Northeast).

Examining the interaction terms in the even-numbered models, political identity again interacted with protest cause in predicting each outcome (b = .251 [p < .001] for police repression in model 2, b = .151 [p = .021] for legal repression in model 4, and b = .220 [p < .001] for punitiveness in model 6); again, the effects were significant at the Bonferroni-adjusted significance level (p < .003) for police repression and punishment and “suggestive” of a relationship for legal repression.

Figure 2 shows predicted values for each outcome in both protest cause conditions. In the BLM condition, greater conservatism (reduced liberalism) was associated with greater support for police and legal repression and greater punitiveness (although the upward trend for legal repression is weak). For respondents assigned to the MAGA condition, greater conservatism (reduced liberalism) was associated with reduced support for police and legal repression and greater punitiveness. At the end points of the continuum, for all outcomes, strong liberals and conservatives preferred more repressive responses to ideological opponents than to ideological allies. Strong liberals again showed more polarized support for police and legal repression than did strong conservatives, while the effects were more symmetrical for punishment preferences.

Figure 2.

Support for repressive responses to protest harm by protest cause (vignette 2 [V2]): (a)support for police repression (V2), (b)support for legal repression (V2), and (c)punishment preferences (V2).

Tests of differences in adjusted predictions again show that strong liberals supported significantly less police repression (difference = −.747, p < .001) and legal repression (difference = −.499, p = .001) of BLM protesters compared with MAGA protesters, while strong conservatives did not support significantly more police (difference = .258, p = .067) or legal repression (difference = .105, p = .481) of BLM protesters compared with MAGA protesters. Both strong liberals (difference = −.452, p = .001) and strong conservatives (difference = .429, p = .002) again supported significantly more punishment of the other side’s protesters.

Three-Way Interactions

We next examine three-way interactions among political identity, protest cause, and incident characteristics, including the police victim manipulations (i.e., political identity × protest cause × police victim in both vignettes) and severity manipulations (i.e., political identity × protest cause × harm in vignette 1 and political identity × protest cause × positive test in vignette 2). Significant interactions would indicate that the degree of partisan bias varies across those conditions.

Results from each model are presented in online Appendix B (Table B1). Overall, there is little evidence partisan bias varies by these situational characteristics: only one of the 12 interactions tested was statistically significant at the Bonferroni-adjusted significance level (p < .003). Specifically, political identity, protest cause, and the police victim manipulation interacted in predicting punitiveness in Vignette 1 (b = .332, p = .001). Figure 3 shows the political identity × protest cause interactions for the police victim and civilian victim conditions in vignette 1. As expected, partisan bias was stronger in the police victim condition than in the civilian victim condition. Strong liberals and strong conservatives recommended more lenient punishments for ideologically aligned protesters and harsher punishments for ideologically opposed protesters in the police condition only.¹⁰

Figure 3.

Support for punishment of protester by protest cause, civilian versus police victim (vignette 1 [V1]): (a)predicted punishment, civilian victim (V1) and (b)predicted punishment, police victim (V1).

Supplemental Analysis: Support for Trump as an Alternative Measure of Partisanship

We also performed supplemental analyses assessing partisan bias on the basis of alignment with President Trump, which may likewise promote perceptions of BLM or MAGA protesters as ideological opponents or allies. The results suggest that partisan bias is also present when Trump voters are compared with non–Trump voters (see online Appendix C, Table C1 and Figure C1). Controlling for identification on a liberal-conservative continuum, support for Trump (i.e., having voted for Trump in 2020) interacted with protest ideology in predicting all outcomes, including support for police repression (b = .441 [p = .001] in vignette 1, b = .773 [p < .001] in vignette 2), legal repression (b = .422 [p = .006] in vignette 1, b = .540 [p = .001] in vignette 2), and punishment (b = .454 [p < .001] in vignette 1, b = .427 [p = .004] in vignette 2). Overall, Trump voters supported greater repression of BLM protests and less repression of MAGA protests, and non–Trump voters supported the opposite pattern.

Discussion

In the present study we explored partisan bias in public support for repressive and punitive government responses to violence or harm occurring at protests (including police repression, legal repression, and punishment of protesters). We used a pair of experimental vignettes to test three hypotheses: that partisans would prefer more repressive responses to ideological opponents, and less repressive responses to ideological allies; that partisan bias would be greater for less serious (i.e., less harmful or intentional) infractions; and that partisan bias would be greater when the recipient of the violence or harm was a police officer rather than a civilian.

The first hypothesis was supported. Overall, respondents’ placement on a liberal-conservative continuum predicted support for different government responses when violence or harm occurred at protests that were more ideologically aligned or more ideologically opposed. That this pattern appeared across multiple scenarios and outcomes, as well as using support for Trump as an alternate measure of partisanship, suggests that the partisan bias we observed was not specific to a particular type of harm or violence, a particular repressive government response, or a particular measure of ideology. These findings align broadly with prior research on partisan bias in perceptions of, and responses to, protest violence (Barker et al. 2021; Edwards and Arnon 2021; Hsiao and Radnitz 2021). Future research should, however, explore why the effects were weaker for legal repression than for police repression or punishment. As noted above, although differences were observed for strong liberals vs. strong conservatives, the interaction terms in the models predicting support for legal repression failed to reach the Bonferroni-adjusted level of significance (p < .003). One possibility (as we discuss below) is that conservatives were less willing than liberals to apply legal bans to either protest.

We also found evidence of asymmetry. As in some prior research (Barker et al. 2021; Hsiao and Radnitz 2021), partisan bias was more pronounced among liberal respondents than among conservative respondents (i.e., strong liberal respondents strongly favored BLM protesters over MAGA protesters, while strong conservatives weakly favored MAGA protesters over BLM protesters). Contributing to this pattern, conservatives had robust support for police repression but relatively low support for legal repression for both protest causes. It may be that conservatives—who tend to be supportive of police intervention broadly (e.g., Silver and Pickett 2015) and who are increasingly protective of free speech in a perceived “cancel culture” (e.g., Fahey, Roberts, and Utych 2023)—focused on the nature of the government intervention rather than the protest cause in evaluating the incidents. Along these lines, regarding support for police repression, conservatives may have been more likely than liberals to believe that the broader social system and its representatives (e.g., police) are just and that the repression of protests by police is justified to maintain social order (Goff et al. 2022; Silver 2020; Silver, Goff, and Iceland 2022). Another possibility is that conservatives were less likely to view police presence as a form of repression, perhaps because policing of right-leaning protests is often less harsh than policing of left-leaning protests (e.g., Jackson 2021). Alternatively, these effects may simply have reflected responses to the specific protest movements manipulated. Future research could further investigate the extent and causes of partisan asymmetry in support of repression, particularly in response to other protest movements.

Partisan bias in punishment preferences was more symmetrical: strong liberals and conservatives both preferred to punish the other side’s protesters more harshly than their own. This finding may reflect a distinction between government actions that affect protest movements and government actions that affect individual protesters. This finding also indicates that—although punishment preferences are typically calibrated to the harmfulness or wrongfulness of the act (e.g., Carlsmith, Darley, and Robinson 2002), people may perceive wrongdoers as more culpable or blameworthy when they are ideological opponents. This possibility aligns with research showing partisan biases in attributions of blame to political elites (Bisgaard 2015) and suggests that people hold similar biases when assigning criminal blame.

In contrast, our second and third hypotheses were generally not supported by the data. That is, partisan bias did not differ significantly across most of the scenarios involving different levels of infraction severity or police versus civilian victims. There was one exception. In vignette 1, partisan bias in criminal punishment was heightened when the victim was a police officer but was absent when the victim was a civilian. Although a single significant result should be interpreted with caution, the finding does align with our theoretical expectations: because police officers are representatives of the state (e.g., Bradford et al. 2014), people may perceive violence against them as symbolic and justifiable if they agree with a protest’s cause (or vice versa) (Maguire et al. 2020). In contrast, violence against bystanders may be seen as a criminal offense and potentially detrimental to a protest movement’s success (e.g., Feinberg et al. 2020), leading to condemnation by opponents and allies alike. However, even adjusting for multiple tests, we stress that this was only one significant finding and should not be overinterpreted.

Taken together, our findings have theoretical and practical implications and suggest additional directions for future research. Theoretically, the results speak to the effects of partisan bias in shaping responses to violent or harmful acts, consistent with the notion that membership in ideological “tribes” shapes how people think about each other (Iyengar and Westwood 2015). As noted above, partisan respondents reading the exact same scenario viewed identical actions by the “other side” as more worthy of repression and punishment than the same actions by “their side.” A possibility is that such divergent responses to the same actions represent moral evaluations of the offender rather than evaluations of the offense (Silver and Berryessa 2023), as political affiliation is often used to infer information about moral values (Bruchmann, Koopmann-Holm, and Scherer 2018). Future research could explore how partisan biases shape perceptions of those who engage in violence or harm, and perceptions of appropriate responses to violence or harm, more broadly.

It was also notable that respondents supported forms of repression that are not legally viable, including the banning of future protests or political action by supporters of a movement (likely a violation of the First Amendment) and criminally punishing an individual for attending a protest while ill (an action that is, although morally questionable, not a crime). One interpretation is that the survey responses did not truly represent people’s views: respondents may engage in “intensity matching” (Kahneman 2011) and respond according to their feelings of approval or disapproval, disregarding the specific response options. However, substantial minorities of respondents did claim to support such measures: 27 percent to 29 percent of respondents somewhat or strongly supported banning future protests, while 30 percent to 34 percent supported banning future action by the political group, across both vignettes, and 23 percent of respondents supported a prison sentence for the individual who attended a protest with COVID-19. Future research might probe whether people believe it is appropriate for the government to suppress speech by ideological opponents or imprison protesters for actions that would not typically be criminally punished.

Our research also has practical implications for advocates who rely on collective action to mobilize public opinion, which is crucial to the success of social movements (Thomas and Louis 2014). If the same protest tactics provoke different responses among partisans, advocates must be sensitive to how actions may be perceived by ideological opponents and allies. We also found that the public has greater support for repressive measures against protests that involve more severe violence and harm as well as violence against police officers, consistent with prior research on violent and nonviolent protest tactics (Feinberg et al. 2020; Hsiao and Radnitz 2021). Policy makers and government actors should also consider how repressive tactics may be perceived by ideological opponents and allies among the public (Novick and Pickett 2022).

With these implications in mind, a few of the main effects, although not the focus of the study, should be noted. Although the severity of the infraction (i.e., the positive test manipulation) in vignette 2 was associated with support for all outcomes, the severity of the infraction (i.e., the injury manipulation) in vignette 1 was only associated with punishment. Perhaps preferences regarding the repression of protest movements are more sensitive to intent than to harm, while punishment preferences incorporate both (Darley 2009). Whether a police officer was a victim was only associated with support for police repression in vignette 1, suggesting that violence against the police may lead to support for harsher policing of protests. That respondents were less supportive of police and legal repression of BLM protests (compared with MAGA protests) but were similarly punitive to individual protesters from each movement further suggests the public views repression of protest movements and the punishment of protesters differently.

Two findings regarding the control variables also suggest a need for future research. Racial resentment was a strong predictor of all outcomes. Although supplemental analyses (described in note 8) showed the effect was significantly moderated by protest cause (i.e., racial resentment was more strongly associated with support for the repression of BLM protests), racial resentment nevertheless predicted greater support for the repression of both BLM and MAGA protests. Given that racial resentment is positively associated with support for the MAGA movement (e.g., Drakulich et al. 2020), this is somewhat surprising. Future research could explore whether racial resentment captures broader preferences for maintaining social hierarchies, resulting in discomfort with protests that by definition challenge the existing social order, or was simply a proxy for unmeasured variables (e.g., system justification or social dominance orientation). Additionally, having participated in more protests was associated with increased support for repression in most of the models, likewise a counterintuitive finding. Future work might consider whether individuals who take part in protests are more attuned to the possible negative consequences of violent or harmful protest actions (including, in the case of COVID-19 transmission, harm to other protesters).

The study has several limitations. The use of vignettes means we cannot be sure of the extent to which our study would reflect responses to real-world protests. Relatedly, the vignettes describe specific scenarios and protests and cannot capture the range of considerations that may drive Americans’ support for protest repression. The use of well-known protest movements to signal ideological ingroup/outgroup membership means that people may have evaluated the scenarios in part by drawing on beliefs about current events (e.g., the January 6 insurrection). As such, the study would have benefited from a control group to assess responses to violence or harm at nonpartisan events (e.g., “convivial gatherings”; see McCarthy, Martin, and McPhail 2007). Another limitation was that both vignettes were presented in the same survey, and their order was not randomized. Therefore, it is plausible that the scenario described in vignette 1 may have influenced respondents’ interpretation of the events in vignette 2.¹¹ For example, by priming respondents to think about violence, the vignette 1 scenario may have predisposed respondents view COVID-19 transmission as punishable act rather than a tragic accident. Regarding the specific content of the vignettes, it is also plausible that respondents may have found characteristics of the short scenarios ambiguous. For example, the bystander in vignette 1 was only described as “observing the protest.” Although we described the bystander as an observer to avoid suggesting participation in the protest, responses to the vignette could potentially have varied if respondents assumed the bystander was a protester or counterprotester. The study also relied on a sample that is not representative except on select demographic measures. Future research could address these limitations by exploring responses to real-life protests and replicating the study using alternate samples and methodologies.

Despite these limitations, the present study provides insight into the public’s preferences for repressive government responses—including police repression, legal repression, and punishment of individual protesters—to protest violence or harm. Overall, those preferences appear to be colored by partisan bias, such that people prefer harsher repressive measures against ideological opponents than ideological allies. More broadly, our results speak to the importance of considering political “tribalism” in understanding the public’s preferences for social control. We urge future researchers to continue this line of work.

Supplemental Material

sj-docx-1-srd-10.1177_23780231231182908 – Supplemental material for Punishing Protesters on the “Other Side”: Partisan Bias in Public Support for Repressive and Punitive Responses to Protest Violence

Supplemental material, sj-docx-1-srd-10.1177_23780231231182908 for Punishing Protesters on the “Other Side”: Partisan Bias in Public Support for Repressive and Punitive Responses to Protest Violence by Jason R. Silver and Luzi Shi in Socius

Supplemental Material

sj-txt-2-srd-10.1177_23780231231182908 – Supplemental material for Punishing Protesters on the “Other Side”: Partisan Bias in Public Support for Repressive and Punitive Responses to Protest Violence

Supplemental material, sj-txt-2-srd-10.1177_23780231231182908 for Punishing Protesters on the “Other Side”: Partisan Bias in Public Support for Repressive and Punitive Responses to Protest Violence by Jason R. Silver and Luzi Shi in Socius

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is funded using internal funds from Rutgers University–Newark and Bridgewater State University.

ORCID iD

Jason R. Silver

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Jason R. Silver (formerly Jasmine R. Silver) is an assistant professor of criminal justice at Rutgers University–Newark. He received his PhD in criminal justice from the University at Albany in 2018. Broadly, his research centers on the role of moral intuition and ideology in shaping responses to crime and criminal justice. His recent work appears in Social Forces, the Journal of Research in Crime and Delinquency, Sexual Abuse, Criminal Justice and Behavior, and the Journal of Experimental Criminology, among other outlets.

Luzi Shi is an assistant professor in the Department of Criminology and Criminal Justice at the University of Rhode Island. She received her PhD in criminal justice from the University at Albany in 2020. Her research interests include public opinion about crime and criminal justice, media, gender, victimology, and inequality. Her work has appeared in Criminology, the Journal of Experimental Criminology, and Sociological Quarterly, among other outlets.

References

Baggetta

Matthew

Myers

Daniel J.

2022. “Interpreting Unrest: How Violence Changes Public Opinions about Social Movements.” Social Movement Studies 21(4):469–492.

Baranauskas

Andrew J.

2022. “News Media and Public Attitudes toward the Protests of 2020: An Examination of the Mediating Role of Perceived Protester Violence.” Criminology & Public Policy 21(1):107–123.

Barkan

Steven

. 2006. “Criminal Prosecution and the Legal Control of Protest.” Mobilization: An International Quarterly 11(2)181–94.

Barker

David

Nalder

Kimberly

Newham

Jessica

. 2021. “Clarifying the Ideological Asymmetry in Public Attitudes toward Political Protest.” American Politics Research 49(2):157–70.

Benjamin

Daniel J.

Berger

James O.

Johannesson

Magnus

Nosek

Brian A.

Wagenmakers

E.-J.

Berk

Richard

Bollen

Kenneth A.

, et al. 2018. “Redefine Statistical Significance.” Nature Human Behaviour 2(1):6–10.

Berinsky

Adam J.

Margolis

Michele F.

Sances

Michael W.

2014. “Separating the Shirkers from the Workers? Making Sure Respondents Pay Attention on Self-Administered Surveys.” American Journal of Political Science 58(3):739–53.

Bisgaard

Martin

. 2015. “Bias Will Find a Way: Economic Perceptions, Attributions of Blame, and Partisan-Motivated Reasoning during Crisis.” Journal of Politics 77(3):849–60.

Blake

Aaron

. 2022. “More Republicans Now Call Jan. 6 a ‘Legitimate Protest’ than a ‘Riot.’” The Washington Post, July 7. Retrieved June 21, 2023. https://www.washingtonpost.com/politics/2022/07/07/many-republicans-no-longer-call-jan-6-an-insurrection-or-even-riot/.

Boas

Taylor C.

Christenson

Dino P.

Glick

David M.

2020. “Recruiting Large Online Samples in the United States and India: Facebook, Mechanical Turk, and Qualtrics.” Political Science Research and Methods 8(2):232–50.

10.

Bradburn

Norman M.

Sudman

Seymour

Wansink

Brian

. 2004. Asking Questions: The Definitive Guide to Questionnaire Design—For Market Research, Political Polls, and Social and Health Questionnaires. Hoboken, NJ: John Wiley.

11.

Bradford

Ben

Murphy

Kristina

Jackson

Jonathan

. 2014. “Officers as Mirrors: Policing, Procedural Justice and the (Re) Production of Social Identity.” British Journal of Criminology 54(4):527–50.

12.

Brantley-Jones

Kiara

. 2021. “False Equivalency between Black Lives Matter and Capitol Siege: Experts, Advocates.” ABC News, January 16. Retrieved June 21, 2023. https://abcnews.go.com/US/false-equivalency-black-lives-matter-capitol-siege-experts/story?id=75251279.

13.

Bruchmann

Kathryn

Koopmann-Holm

Birgit

Scherer

Aaron

. 2018. “Seeing Beyond Political Affiliations: The Mediating Role of Perceived Moral Foundations on the Partisan Similarity-Liking Effect.” PLoS ONE 13(8):e0202101.

14.

Brutger

Ryan

Kertzer

Joshua D.

Renshon

Jonathan

Tingley

Dustin

Weiss

Chagai M.

2022. “Abstraction and Detail in Experimental Design.” American Journal of Political Science. doi:10.1111/ajps.12710.

15.

Bugden

Dylan

. 2020. “Does Climate Protest Work? Partisanship, Protest, and Sentiment Pools.” Socius: Sociological Research for a Dynamic World 6. doi:10.1177/2378023120925949.

16.

Carlsmith

Kevin M.

Darley

John M.

Robinson

Paul H.

2002. “Why Do We Punish? Deterrence and Just Deserts as Motives for Punishment.” Journal of Personality and Social Psychology 83(2):284–99.

17.

Chandler

Jesse

Rosenzweig

Cheskie

Moss

Aaron J.

Robinson

Jonathan

Litman

Leib

. 2019. “Online Panels in Social Science Research: Expanding Sampling Methods Beyond Mechanical Turk.” Behavior Research Methods 51:2022–38.

18.

Chenoweth

Erica

Pressman

Jeremy

. 2020. “This Summer’s Black Lives Matter Protesters Were Overwhelmingly Peaceful, Our Research Finds.” The Washington Post, September 4. Retrieved June 21, 2023. https://www.washingtonpost.com/politics/2020/10/16/this-summers-black-lives-matter-protesters-were-overwhelming-peaceful-our-research-finds/.

19.

Chenoweth

Erica

Stephan

Maria J.

Stephan

Maria

. 2011. Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict. New York: Columbia University Press.

20.

Cho

Ahram

Cooper

Maisha N.

Updegrove

Alexander H.

Luo

Fei

. 2022. “How Representative of the Protest Movement against Police Violence Are Opportunistic People Looking to Commit Crime? Racial Differences in Public Opinion.” Race and Justice. doi:10.1177/21533687221117278.

21.

Cobbina

Jennifer E.

2019. Hands Up, Don’t Shoot: Why the Protests in Ferguson and Baltimore Matter, and How They Changed America. New York: New York University Press.

22.

Cobbina

Jennifer

LaCourse

Ashleigh

Brooke

Erika J.

Chaudhuri

Soma

. 2021. “Protesting during a Pandemic: Narratives on Risk Taking and Motivation to Participating in the 2020 March on Washington.” Crime & Delinquency 67(8):1195–220.

23.

Darley

John M.

2009. “Morality in the Law: The Psychological Foundations of Citizens’ Desires to Punish Transgressions.” Annual Review of Law and Social Science 5:1–23.

24.

Davenport

Christian

. 2007. “State Repression and Political Order.” Annual Review of Political Science 10:1–23.

25.

Ditto

Peter H.

Liu

Brittany S.

Clark

Cory J.

Wojcik

Sean P.

Chen

Eric E.

Grady

Rebecca H.

Celniker

Jared B.

, et al. 2019. “At Least Bias is Bipartisan: A Meta-analytic Comparison of Partisan Bias in Liberals and Conservatives.” Perspectives on Psychological Science 14(2):273–91.

26.

Drakulich

Kevin

Denver

Megan

. 2022. “The Partisans and the Persuadables: Public Views of Black Lives Matter and the 2020 Protests.” Perspectives on Politics 20(4):1191–1208.

27.

Drakulich

Kevin

Wozniak

Kevin H.

Hagan

John

Johnson

Devon

. 2020. “Race and Policing in the 2016 Presidential Election: Black Lives Matter, the Police, and Dog Whistle Politics.” Criminology 58(2):370–402.

28.

Earl

Jennifer

. 2011. “Protest Arrests and Future Protest Participation: The 2004 Republican National Convention Arrestees and the Effects of Repression.” Pp. 141–73 in Special Issue Social Movements/Legal Possibilities, Vol. 54, edited by A.

Sarat.

Bingley, UK: Emerald Group.

29.

Edwards

Pearce

Arnon

Daniel

. 2021. “Violence on Many Sides: Framing Effects on Protest and Support for Repression.” British Journal of Political Science 51(2):488–506.

30.

Enns

Peter K.

Ramirez

Mark D.

2018. “Privatizing Punishment: Testing Theories of Public Support for Private Prison and Immigration Detention Facilities.” Criminology 56(3):546–73.

31.

Fahey

James J.

Roberts

Damon C.

Utych

Stephen M.

2023. “Principled or Partisan? The Effect of Cancel Culture Framings on Support for Free Speech.” American Politics Research 51(1):69–75.

32.

Feinberg

Matthew

Willer

Robb

Kovacheff

Chloe

. 2020. “The Activist’s Dilemma: Extreme Protest Actions Reduce Popular Support for Social Movements.” Journal of Personality and Social Psychology 119(5):1086–111.

33.

Fowler

Anthony

Hill

Seth J.

Lewis

Jeffrey B.

Tausanovitch

Chris

Vavreck

Lynn

Warshaw

Christopher

. 2023. “Moderates.” American Political Science Review 117(2):643–60.

34.

Freedman

David A.

2008. “On Regression Adjustments to Experimental Data.” Advances in Applied Mathematics 40(2):180–93.

35.

Giersch

Jason

. 2019. “Punishing Campus Protesters Based on Ideology.” Research & Politics 6(4):2053168019892129.

36.

Goff

Kerby

Silver

Eric

Iceland

John

. 2022. “The Resonance of Repression: Moral Intuitions, Skepticism toward Racial Injustice, and Public Support for Trump’s ‘Law and Order’ Response to the 2020 Racial Justice Protests.” Socius: Sociological Research for a Dynamic World 8. doi:10.1177/23780231221110277.

37.

Gutting

Raynee Sarah

. 2020. “Contentious Activities, Disrespectful Protesters: Effect of Protest Context on Protest Support and Mobilization across Ideology and Authoritarianism.” Political Behavior 42(3):865–90.

38.

Heaney

Michael

Rojas

Fabio

. 2011. “The Partisan Dynamics of Contention: Demobilization of the Antiwar Movement in the United States, 2007–2009.” Mobilization: An International Quarterly 16(1):45–64.

39.

Horowitz

Juliana Menasce

. 2021. “Support for Black Lives Matter Declined after George Floyd Protests, but Has Remained Unchanged Since.” Pew Research Center, September 27. Retrieved June 21, 2023. https://www.pewresearch.org/fact-tank/2021/09/27/support-for-black-lives-matter-declined-after-george-floyd-protests-but-has-remained-unchanged-since/.

40.

Hsiao

Yuan

Radnitz

Scott

. 2021. “Allies or Agitators? How Partisan Identity Shapes Public Opinion about Violent or Nonviolent Protests.” Political Communication 38(4):479–97.

41.

Huff

Connor

Kruszewska

Dominika

. 2016. “Banners, Barricades, and Bombs: The Tactical Choices of Social Movements and Public Opinion.” Comparative Political Studies 49(13):1774–1808.

42.

Irvine

Chris

. 2020. “George Floyd Protests: Rioters Target Police across US; 4 Shot in St. Louis, 1 in Vegas, Bronx Hit-and-Run Caught on Video.” Fox News, June 2. Retrieved June 21, 2023. https://www.foxnews.com/us/george-floyd-protests-rioters-police-shot-hit-and-run.

43.

Iyengar

Shanto

Westwood

Sean J.

2015. “Fear and Loathing across Party Lines: New Evidence on Group Polarization.” American Journal of Political Science 59(3):690–707.

44.

Jackson

Kelly Carter

. 2021. “The Inaction of Capitol Police Was by Design.” The Atlantic, January 8. Retrieved June 21, 2023. https://www.theatlantic.com/culture/archive/2021/01/inaction-capitol-police-was-design/617590/.

45.

Kahan

Dan M.

Peters

Ellen

Dawson

Erica Cantrell

Slovic

Paul

. 2017. “Motivated Numeracy and Enlightened Self-Government.” Behavioural Public Policy 1(1):54–86.

46.

Kahneman

2011. Thinking, Fast and Slow. New York: Macmillan.

47.

Kishi

Roudabeh

Jones

Sam

. 2020. “Demonstrations and Political Violence in America: New Data for Summer 2020.” Armed Conflict Location & Event Data Project, September 3. Retrieved June 21, 2023. https://acleddata.com/2020/09/03/demonstrations-political-violence-in-america-new-data-for-summer-2020/.

48.

Kishi

Roudabeh

Stall

Hampton

Jones

Sam

. 2020. “The Future of ‘Stop the Steal’: Post-election Trajectories for Right-Wing Mobilization in the United States.” Armed Conflict Location & Event Data Project, December 10. Retrieved June 21, 2023. https://acleddata.com/2020/12/10/the-future-of-stop-the-steal-post-election-trajectories-for-right-wing-mobilization-in-the-us/.

49.

Kishi

Roudabeh

Stall

Hampton

Wolfson

Aaron

Jones

Sam

. 2021. “A Year of Racial Justice Protests: Key Trends in Demonstrations Supporting the BLM Movement.” Armed Conflict Location & Event Data Project, May 25. Retrieved June 21, 2023. https://acleddata.com/2021/05/25/a-year-of-racial-justice-protests-key-trends-in-demonstrations-supporting-the-blm-movement/.

50.

Maguire

Edward

Barak

Maya

Wells

William

Katz

Charles

. 2020. “Attitudes towards the Use of Violence against Police among Occupy Wall Street Protesters.” Policing: A Journal of Policy and Practice 14(4):883–99.

51.

McCarthy

John D.

Martin

Andrew

McPhail

Clark

. 2007. “Policing Disorderly Campus Protests and Convivial Gatherings: The Interaction of Threat, Social Organization, and First Amendment Guarantees.” Social Problems 54(3):274–96.

52.

Metcalfe

Christi

Pickett

Justin T.

2022. “Public Fear of Protesters and Support for Protest Policing: An Experimental Test of Two Theoretical Models.” Criminology 60(1):60–89.

53.

Mische

Ann

. 2009. Partisan Publics. Princeton, NJ: Princeton University Press.

54.

Novick

Rachel

Pickett

Justin T.

2022. “Black Lives Matter, Protest Policing, and Voter Support for Police Reform in Portland, Oregon.” Race and Justice. doi:10.1177/21533687221117281.

55.

Olzak

Susan

. 2021. “Does Protest against Police Violence Matter? Evidence from US Cities, 1990 through 2019.” American Sociological Review 86(6):1066–99.

56.

Shuman

Eric

Hasan-Aslih

Siwar

Zomeren

Martijn van

Saguy

Tamar

Halperin

Eran

. 2022. “Protest Movements Involving Limited Violence Can Sometimes Be Effective: Evidence from the 2020 BlackLivesMatter Protests.” Proceedings of the National Academy of Sciences 119(14):e2118990119.

57.

Silver

Eric

Goff

Kerby

Iceland

John

. 2022. “Social Order and Social Justice: Moral Intuitions, Systemic Racism Beliefs, and Americans’ Divergent Attitudes toward Black Lives Matter and Police.” Criminology 60(2):342–69.

58.

Silver

Jasmine R.

2020. “Moral Motives, Police Legitimacy and Acceptance of Force.” Policing: An International Journal 43(5):799–815.

59.

Silver

Jason R.

Berryessa

Colleen M.

2023. “Remorse, Perceived Offender Immorality, and Lay Sentencing Preferences.” Journal of Experimental Criminology 19:425–63.

60.

Silver

Jason R.

Pickett

Justin T.

2015. “Toward a Better Understanding of Politicized Policing Attitudes: Conflicted Conservatism and Support for Police Use of Force.” Criminology 53(4):650–76.

61.

Simpson

Brent

Willer

Robb

Feinberg

Matthew

. 2018. “Does Violent Protest Backfire? Testing a Theory of Public Reactions to Activist Violence.” Socius: Sociological Research for a Dynamic World 4. doi:10.1177/2378023118803189.

62.

Slobin

Sara

Hart

Sam

. 2021. “When the Right Wing Rallies.” Reuters, April 15. Retrieved June 22, 2023. https://graphics.reuters.com/USA-CAPITOL/SECURITY/xegpbxoadpq/.

63.

StataCorp. 2021. Stata 17 Base Reference Manual. College Station, TX: Stata Press.

64.

Stephan

Maria J.

Chenoweth

Erica

. 2008. “Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict.” International Security 33(1):7–44.

65.

Thomas

Emma F.

Louis

Winnifred R.

2014. “When Will Collective Action Be Effective? Violent and Non-violent Protests Differentially Influence Perceptions of Legitimacy and Efficacy among Sympathizers.” Personality and Social Psychology Bulletin 40(2):263–76.

66.

Tilly

Charles

. 1978. From Mobilization to Revolution. New York: McGraw-Hill.

67.

Tyler

David H.

Barak

Maya

Maguire

Edward R.

Wells

William

. 2018. “The Effects of Procedural Injustice on the Use of Violence against Police by Occupy Wall Street Protesters.” Police Practice and Research 19(2):138–52.

68.

U.S. Census Bureau. 2022. “U.S. Census Bureau Quick Facts.” Retrieved June 22, 2023. https://www.census.gov/quickfacts/fact/table/US/PST045222.

69.

Zack

Elizabeth S.

Kennedy

John

Long

J. Scott

. 2019. “Can Nonprobability Samples Be Used for Social Science Research? A Cautionary Tale.” Survey Research Methods 13(2):215–27.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.18 MB

0.03 MB